This is happening to our most important box on campus so we are no too
eager to try it again. That box is now running its backup via remote tar.
It happens on every backup attempt via NFS. Next time I have some downtime
on the box I will try again and might be able to give you more details.
The goto loop in nfs_read_inode seems not to have a very clean exit
condition. Should it not fail completely after a couple of retries?
On 28 Apr 1999, Trond Myklebust wrote:
> Christoph Lameter <christoph@lameter.com> writes:
>
> > We do regular backups each night having all boxes NFS mounted on a central
> > server equipped with a tape drive. Every Tuesday night the box rebooted
> > since we switched to 2.2.X. I finally ran the backup for that special
> > machine manually.
> >
> > Backup ran normally for awhile until it got to the /dev directory. Then
> > there was an endless scroll of barely readable NFS error messages on the
> > console.
> >
> > Something about "NFS client: inode locked/unavailable" and "retrying".
> > This went on for awhile until some system resources were exhausted and the
> > watchdog daemon took the machine down.
> >
> > I could not find any messages at all in the system logs.
> >
> > Any hints on how to proceed with that one? Backing up other linux boxes
> > via NFS works just fine.
>
> I can't find anything which corresponds to that message in the
> kernel.
>
> If the error is repeatable, do you think you could turn on the NFS
> debugging: probably 'echo 1 >/proc/sys/sunrpc/nfs_debug' (turns on the
> VFS == fs/nfs/inode.c debugging) should suffice?
-----------------------------------------------------------------------------
Christoph Lameter, MSCS, M.Div.
Available for a job or consulting (see http://lameter.com/consulting.html)
-----------------------------------------------------------------------------
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/